Evaluation models for soil nutrient based on support vector machine and artificial neural networks.

Area of Science:

• Agricultural Science
• Environmental Science
• Data Science

Background:

• Soil fertility and environmental impact are significantly influenced by soil nutrient content.
• Traditional soil nutrient evaluation methods are complex and difficult to implement in practical settings.
• Developing efficient and accurate soil nutrient assessment tools is crucial for sustainable agriculture.

Purpose of the Study:

• To develop and compare comprehensive evaluation models for soil nutrient content.
• To assess the effectiveness of Support Vector Machine (SVM), Multiple Linear Regression (MLR), and Artificial Neural Networks (ANNs) in soil nutrient evaluation.
• To identify the most accurate machine learning models for practical soil nutrient assessment.

Main Methods:

• Utilized Support Vector Machine (SVM), Multiple Linear Regression (MLR), and Artificial Neural Networks (ANNs) for model development.
• Employed soil organic matter, total nitrogen, alkali-hydrolysable nitrogen, rapidly available phosphorus, and rapidly available potassium as independent variables.
• Soil nutrient content evaluation level served as the dependent variable in the models.

Main Results:

• Support Vector Machine (SVM) models achieved average prediction accuracies of 77.87% and 83.00%.
• The General Regression Neural Network (GRNN) model demonstrated a high average prediction accuracy of 92.86%.
• Both SVM and GRNN models proved effective in assessing soil nutrient levels.

Conclusions:

• Machine learning models, particularly SVM and GRNN, offer effective solutions for soil nutrient evaluation.
• These models provide a more practical and accurate alternative to traditional soil testing methods.
• The findings support the application of SVM and GRNN for determining soil nutrient status in agricultural practices.